In this design guide, we describe the configuration and integration of PowerEdge R740 and PowerEdge R740xd servers in an environment that can use vGPUs to improve MLOps for model training and inference. These PowerEdge servers are a 2U, two-socket platform with scalability and performance that you can adapt for various applications. They support 2nd Gen Intel Xeon Scalable processors (previously code-named Cascade Lake). The PowerEdge servers can use up to 24 NVMe drives, or a total of 32 x 2.5-inch or 18 x 3.5-inch drives. Embedded intelligence and automation are provided through iDRAC9. The OpenManage portfolio simplifies the IT life cycle from deployment to retirement.
PowerEdge R740 and PowerEdge R740xd servers support up to three A100 GPUs. When you install an additional network adapter (such as a ConnectX SmartNIC) in the server, the PowerEdge R740 or PowerEdge R740xd server can support up to two A100 GPUs. Both these PowerEdge server models are NVIDIA-Certified Systems.
You can choose vSAN Ready Nodes for preconfigured, tested, and certified PowerEdge servers that run VMware vSAN. For more information, see . You can also deploy for a fully integrated solution and turnkey experience. VxRail V570 and VxRail V570F models support NVIDIA GPUs. Contact your Dell Technologies Sales Representative for details about the appropriate VxRail models and options that meet the requirements of this design guide.
Dell EMC PowerScale storage helps unlock the structure within data and address the challenges of unstructured data management. PowerScale is the next evolution of OneFS—the operating system powering the industry’s leading scale-out NAS platform. The PowerScale family includes Dell EMC Isilon nodes and PowerScale nodes, with PowerScale OneFS running across all of them. The software-defined architecture of OneFS provides simplicity at scale, intelligent insights, and the ability to have any data anywhere it needs to be. Whether hosting file shares or home directories, or delivering high-performance data access for applications such as analytics, video rendering and life sciences, PowerScale can seamlessly scale performance, capacity, and efficiency to handle any unstructured data workload. The new PowerScale all-flash platforms co-exist seamlessly in the same cluster with your existing Isilon nodes to drive your traditional and modern applications.
In this design guide, we use PowerScale as storage for the data lake—the data repository for unstructured data that you can use for neural network training. Isilon F800 All-Flash Scale-out NAS storage is ideal storage, delivering the analytics performance and extreme concurrency at scale to consistently feed the most data hungry deep learning algorithms.
In this design guide, we use the Dell EMC PowerSwitch S5232F-ON switch for 100 GbE network connectivity. Using the S5332F-ON switch, you can build high-performance, cost-efficient data center leaf/spine fabric featuring 32 x 100 GbE QSFP28 ports. We use the S5248F-ON switch for 25 GbE network connectivity. It features 48 x 25 GbE SFP28 ports, 4 x 100 GbE QSFP28 ports, and 2 x 100 GbE QFSP28-DD ports. The S5232F-ON and S5248F-ON switches support Open Network Install Environment (ONIE) for zero-touch installation of network operating systems.
We use the S4148T switch for 1 GbE out-of-band (OOB) connectivity.